Hi Pal,

On Mon, Oct 3, 2011 at 10:07 PM, Mezei Pál <[email protected]> wrote:
> Can anyone explain what different fields mean in the texturesystem
> statistics output? (I'm using Arnold 3.3.12 with oiio1000) I have searched
> for a detailed explanation in the pdf, but found nothing useful (or I was
> looking at the wrong place)

Without reading the code again in detail to check, here's my
understanding of the various statistics:

> 00:02:48  106mb         | OpenImageIO Texture statistics (0000000002AD2F00, 
> cache = 0000000002A14AD0)

Addresses of the texture system and cache

> 00:02:48  106mb         |   Queries/batches :
> 00:02:48  106mb         |     texture     :  69709758 queries in 69709758 
> batches
> 00:02:48  106mb         |     texture 3d  :  0 queries in 0 batches
> 00:02:48  106mb         |     shadow      :  0 queries in 0 batches
> 00:02:48  106mb         |     environment :  0 queries in 0 batches

Total number of texture lookups of the various types.  (You can ignore
batches, it's not relevant to renderers using OSL as the shading
language, only renderers which have SIMD shading systems.)

> 00:02:48  106mb         |   Interpolations :
> 00:02:48  106mb         |     closest  : 10644669
> 00:02:48  106mb         |     bilinear : 21018019
> 00:02:48  106mb         |     bicubic  : 372800217

Each texture lookup may do a number of interpolations depending on the
filtering settings and anisotropy of the filter footprint.  The sum of
these numbers should be strictly larger than the number of texture
lookups.

> 00:02:48  106mb         |   Average anisotropy : 5.61

Average of the filter anisotropies encountered (should be just
length/width, but I didn't check).  Anisotropic filtering is typically
important when rays hit a surface at a glancing angle, though it can
also happen in other circumstances.  Large anisotropies are expensive
because the filter footprint gets stretched and any prefiltering
(ie, mipmapping) becomes less effective.

> 00:02:48  106mb         |   Max anisotropy in the wild : 1e+006

Maximum anisotropy of the filter footprint which was actually
encountered.

Cheers,
~Chris
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to